xml transformation on the fly?

[Alex]

We am facing a problem with transforming docbook xml to html. I am running
PHP4 on apache and i first tried to use sablotron which took 40+ seconds to
convert a 50kb xml file. When using xsltproc it went down to 10-15. The
machine we have is only 400 mhz but we'd need 2-3 mhz to get anywhere near
real time if we use same method as we are now. I may be missing some
features in xsltlib which can maybe preload the xslt tree or maybe there is
a better processor out threre. I would really appreciate any suggestions on
how to speed this up.

Alex

 

[Nick Kew]

I would really appreciate any suggestions on
how to speed this up.

http://www.outoforder.cc/projects/apache/mod_transform/

 

[Peter Flynn]

We am facing a problem with transforming docbook xml to html. I am running
PHP4 on apache and i first tried to use sablotron which took 40+ seconds
to convert a 50kb xml file. When using xsltproc it went down to 10-15. The
machine we have is only 400 mhz but we'd need 2-3 mhz to get anywhere near
real time if we use same method as we are now.

Real-time XML --> HTML using XSLT needs both speed and memory.

AxKit is the fastest I have seen, but it doesn't scale well to large files
(but 50Kb is very small). However, I have had serious problems getting it
running on some platforms as the authors don't have extensive platform
access for development testing.

Cocoon is the software of choice for many such applications, as it's fairly
fast, and it has a cache for re-serving unaltered already-converted docs.
But it's Java, so it's a memory hog, and its garbage collection is poor.
I've had it running quite happily on a 400MHz machine provided I don't try
to use the machine as my workstation at the same time :-( but do fill it to
the max with memory.

PropelX is a commercial transformation server. I don't have speed metrics,
but ask the supplier (www.propylon.com).

Do you really need real-time transformation? If the documents only change
every few hours, maybe a static conversion using Saxon or similar is a
better answer, triggered each time a document is re-saved.

///Peter